A data-driven approach to generate pseudo-reaction sequences for the thermal conversion of Athabasca bitumen

Abstract

This work focuses on the application of self-modeling multivariate curve resolution (SMCR) methods on the Fourier transform infrared (FTIR) spectra of the liquid products obtained from the thermal cracking of Athabasca bitumen in the temperature range of 300–420 °C and reaction times ranging from 15 min to 27 h. The objective was to develop a reaction pathway for the thermal cracking process from the SMCR methods and to identify key elements of the reaction chemistry that also affected physical properties like viscosity. An important aspect of this work was that minimum external chemical knowledge was used for the chemometric techniques. The SMCR method employed in our study was applied on both temperature-specific and augmented datasets considering all temperatures together to extract resolved concentration and spectral profiles using the alternating least-squares (ALS) optimization. The improvements of particle swarm optimization (PSO) over ALS were investigated with regards to resolution quality, convergence speed, residuals and explained variance. The thermal conversion of Athabasca bitumen was shown to observe a series reaction sequence with methyl transfer dominant at lower temperatures and a greater extent of cracking at higher temperatures along with the formation of lighter products with a higher fraction of mono-substituted aromatics.